OUP user menu

Prokaryotic biodiversity and activity in the deep subseafloor biosphere

John C. Fry , R. John Parkes , Barry A. Cragg , Andrew J. Weightman , Gordon Webster
DOI: http://dx.doi.org/10.1111/j.1574-6941.2008.00566.x 181-196 First published online: 1 November 2008

Abstract

The deep subseafloor biosphere supports a diverse population of prokaryotes belonging to the Bacteria and Archaea. Most of the taxonomic groups identified by molecular methods contain mainly uncultured phylotypes. Despite this several cultured strains have been isolated from this habitat, but they probably do not represent the majority of the population. Evidence is starting to suggest that some of the activities measured, such as sulphate reduction and methanogenesis, reflected in geochemical profiles, are carried out by a small subset of the community detected by molecular methods. It is further possible that heterotrophy may be the most important mode of metabolism in subsurface sediments and heterotrophic microorganisms could dominate the uncultured prokaryotic population. Although, heterotrophy is limited by the increasing recalcitrance of organic matter with depth, this may be counteracted by thermal activation of buried organic matter providing additional substrates at depth.

Keywords
  • prokaryotes
  • activity
  • biodiversity
  • community composition
  • deep biosphere
  • marine subsurface sediments

Introduction

Oceans cover c. 70% of the Earth's surface, and between 5 and 10 billion tons of particulate organic matter is constantly sinking within them and accumulating as marine sediments. This then results in particulate organic matter being concentrated 10 000–100 000-fold higher in sediments, than in overlying seawaters. However, the vast majority of this organic matter is removed by near-surface microbial activity, but over geological time the remainder accumulates and results in the largest global reservoir of organic carbon (15 000 × 1018 g C). Marine sediments have an average depth of around 500 m, but can be up to 10 km in depth with an average overlying water depth of 3800 m. Such water depths produce an average hydrostatic pressure of 380 bar and at maximum water depths (11 km, Mariana Trench) can be up to 1100 bar, which is increased further by the lithostatic pressure of the subsurface sediments. This high pressure combined with the very low near surface sediment temperatures (c. 2 °C), decreasing porosity with increasing depth, and the fact that a large part of the seabed (95%) is at water depths where light intensity is too low to sustain photosynthesis led to the belief that the seabed was flat, biologically inactive and that seafloor sediments were insignificant in terms of their microbial activity (Jannasch et al., 1971; Jannasch & Wirsen, 1973). However, it has since been discovered that the seafloor environment is a dynamic geo- and biosphere that provides a diverse range of living conditions that are host to rich microbial communities (see Jørgensen & Boetius, 2007).

In the 1980s, the first recognition of metabolically active microorganisms in deeply buried sediments arose from studies of pore-water chemistry and the use of radiotracers in deep sediment cores obtained by drilling down to 167 metres below seafloor (mbsf). Potential prokaryotic sulphate reduction and methanogenic activity were detected in buried sediments from three Deep Sea Drilling Project cruises sampling coastal North American sediments (Oremland et al., 1982; Whelan et al., 1986; Tarafa et al., 1987). However, positive results were patchy and data was inconclusive. It was early in the next decade that the first comprehensive depth profiles of microbial activity, total and viable prokaryotic numbers and estimates of cultured biodiversity were published (Cragg et al., 1990; Parkes et al., 1990) showing clear links between activity and the availability of organic carbon and terminal electron acceptors (Cragg et al., 1992). In 1994, a model was formulated for the logarithmic decline of total prokaryotic cell numbers with sediment depth (Parkes et al., 1994), which has now become well accepted (Whitman et al., 1998; Parkes et al., 2000; Jørgensen et al., 2006), and the first culture-independent molecular study to describe biodiversity in the deep marine subsurface was reported (Rochelle et al., 1994).

During the 1990s geologists, geochemists and microbiologists began to realize the importance of the deep marine subsurface habitat, when it was estimated that the deep subseafloor biosphere comprised one-tenth to one-third of the Earth's total biomass and the majority (c. 65%) of the global prokaryotic biomass (Parkes et al., 1994; Whitman et al., 1998). Since then intact cells (Parkes et al., 2000) and intact membrane lipids (Zink et al., 2003; Biddle et al., 2006) have been consistently found in sediments<800 mbsf (Parkes et al., 2000) and recently in sediments down to 1626 mbsf (Roussel et al., 2008). It was on this foundation that the first microbiology focused Ocean Drilling Program (ODP) cruise (Leg 201 in 2002) was planned with the aim to sample both highly productive sites in the Peru Margin and low activity sediments of the Equatorial Pacific and Peru Basin. ODP Leg 201 has now contributed more to our understanding of prokaryotes in the deep marine subsurface than any other ocean drilling expedition to date (for review see Jørgensen et al., 2006).

The aim of this review is to provide a summary of current knowledge of prokaryotic activity and biodiversity in subsurface marine sediments. We will call this habitat the deep subseafloor biosphere, after a similar term coined by Parkes (1994). We define this habitat as the marine subsurface that occurs at sediment depths below 1 mbsf, and will concentrate on studies from ODP cruises, but will include results from other investigations which have aimed to study deep sediment cores (e.g. Reed et al., 2002; Inagaki et al., 2003; Wilms et al., 2007), rather than surface sediments. We will also explore how changes in biodiversity might influence activity and how both are influenced by the geological and geochemical features of the environment. Lastly, we will indicate some future directions for deep marine biosphere research.

Prokaryotic biodiversity

Methods

As with other habitats, the diversity of prokaryotes found in the subseafloor biosphere by cultivation-independent molecular methods is much greater than obtained by standard laboratory culture. Furthermore, the taxa obtained by cultivation methods so far represent only a very small and unrepresentative subset of those revealed by molecular methodologies (e.g. D'Hondt et al., 2004; Toffin et al., 2004). This is thought to be largely as a result of the observed low cultivability (generally <0.1%) of deep biosphere bacteria (e.g. Cragg et al., 1992; Wellsbury et al., 2002; Engelen et al., 2008). Thus, most biodiversity studies have used molecular methods, involving direct extraction of nucleic acids from sediments and PCR amplification of 16S rRNA genes (Giovannoni et al., 1990) and/or functional genes indicative of key anaerobic sedimentary processes [e.g. methanogenesis (mcrA) and sulphate reduction (dsrA)]. These amplified genes are then analysed for diversity by either the construction of gene libraries or by more rapid profiling methods such as denaturing gradient gel electrophoresis (DGGE; Muyzer et al., 1993).

However, molecular methodologies have been especially difficult to use in the study of deep marine sediments because of coextractable interfering substances like humic and fulvic acids, as well as the problem of low prokaryotic cell numbers and consequently low concentrations of extractable nucleic acids. Especially if we consider that the average prokaryotic cell contains <4 fg DNA per cell (Christensen et al., 1995; Abulencia et al., 2006) and subsurface sediments typically have 106–107 cells cm−3 (Parkes et al., 1994, 2000). Such low biomass samples have posed considerable problems for the reliable amplification of 16S rRNA genes from subsurface Bacteria. Because low biomass samples are susceptible to PCR bias by random amplification (Chandler et al., 1997) and in addition most commercial thermostable DNA polymerases and other reagents used in PCR amplification and DNA extraction are often contaminated with small amounts of exogenous bacterial DNA (Rochelle et al., 1992b; Kormas et al., 2003; Webster et al., 2003). This has resulted in the development of a wide variety of specialist methods which include carefully optimized DNA extraction protocols, improved sensitivity using nested PCR and the use of very stringent anticontamination controls (e.g. Rochelle et al., 1992a; Reed et al., 2002; Webster et al., 2003; Sørensen et al., 2004) to ensure that sequences retrieved are representative of subsurface prokaryotes.

It is also important, before subsampling of sediment cores for microbiological analyses, to ensure that sediment samples are of sufficient quality and uncontaminated with drilling fluids, such as seawater, to warrant further time and effort for subsequent DNA extraction. Hence, it has now become a routine procedure for deep subsurface drilling to use a combination of a water soluble chemical tracer and fluorescent microspheres to mimic penetration of bacterial sized particles to monitor possible contamination from seawater and drilling disturbance (Smith et al., 2000; House et al., 2003; Lever et al., 2006).

Community composition of Bacteria

Analysis of 16S rRNA gene libraries has shown that there is substantial diversity of Bacteria in deep marine subsurface sediments, both within and between the phyla. The compositions of the bacterial populations found in a number of studies on the deep subseafloor biosphere are summarized in Fig. 1. These data comprise 1909 clones from 34 16S rRNA gene libraries described in eight different studies, covering a wide range of sediment depths (1–503 mbsf). The 16S rRNA gene libraries are mainly from deep continental margin sites bordering the Pacific Ocean; unfortunately no comparable bacterial 16S rRNA gene libraries from other oceans have yet been obtained. It should also be noted that these libraries were constructed using different molecular approaches (e.g. from different nucleic acid extraction protocols and/or using different domain-specific 16S rRNA gene PCR primers). Therefore, direct quantitative comparison of different clone libraries cannot be made due to the different methodological biases in each case, although qualitative assessments of prokaryotic diversity are possible.

1

Community composition of major taxonomic groups of Bacteria from 16S rRNA gene libraries at various sites and depths in the deep subseafloor biosphere. Each bar shows the relative distribution of major taxonomic groups reported within the clone library. Where practicable all the sediment depths (given as meters below sea floor=mbsf) are reported and illustrated for each site examined. However, when there were less than about 20 clones in a library, the libraries from adjacent depths are combined, and when clone libraries from more than four depths were reported for a site, three to four depths were chosen at intervals to illustrate the range of diversity reported for the site concerned. Some depths are estimated (e.g. c. 30 mbsf) from diagrams provided within the cited paper. The site details and depths for the clone libraries are given to the left of the bars followed by (citation code; number of clones in library). The citation codes are: A, Rochelle (1994); B, Marchesi (2001); C, Reed (2002); D, Inagaki (2003); E, Kormas (2003); F, Newberry (2004); G, Parkes (2005); Webster (2006a); H Inagaki (2006). ODP=Ocean Drilling program. Accepted names are used for the taxonomic groups used (namely Classes of Proteobacteria, phyla and candidate divisions); NT-B2 and NT-B6 are deep branching phylum level groups first named by Reed (2002) and used later by others (NT-B6, e.g. Inagaki et al., 2006; NT-B2, e.g. Webster et al., 2006a).

It is clear that although the bacterial 16S rRNA gene libraries indicate a broadly diverse bacterial population, there is great variability in its composition (Fig. 1). For example, although Proteobacteria averaged 37.4% of clones, and they almost completely dominated the Cascadia Margin ODP site 889/890 (North end, off Vancouver Island; Marchesi et al., 2001) and volcanic ash sediment layers of the Sea of Okhotsk (Inagaki et al., 2003), they were almost absent from Cascadia Margin ODP sites 1244/5 and 1251 (South end, near Hydrate Ridge, off Oregon; Inagaki et al., 2006) and from the Peru Margin ODP site 1230 (Inagaki et al., 2006). Overall, the most abundant subseafloor bacterial groups are Gammaproteobacteria, Chloroflexi and members of the candidate division JS1 (Webster et al., 2004), which make up 18.9%, 17.3% and 26.1% (sum=62.3%; present in 62–70% of the libraries) of the clones, respectively, and so are the dominating groups of Bacteria in this deep sediment habitat. However, the Alpha-, Beta-, Delta- and Epsilonproteobacteria are also found (in 3–50% of the libraries), but are not so common, averaging only 7.8%, 4.9%, 3.7% and 2.1% of the clones, respectively. Of the remaining 21.3% of clones the Planctomycetes are notably abundant (2–26%) at some Peru and Cascadia Margin sites and depths (Parkes et al., 2005; Inagaki et al., 2006), as are the novel groups NT-B2 and NT-B6, both originally found in the Nankai Forearc Basin (Reed et al., 2002). Although the Chloroflexi and the JS1 groups often co-occur, they can also dominate sites with one group being much less abundant. For example, Cascadia Margin sites 1224/5 and 1251, Peru Margin site 1230, Nankai Trough site 1173 and Sea of Okhotsk pelagic clay layers (below 22 mbsf) are dominated by JS1 (Inagaki et al., 2003, 2006; Newberry et al., 2004), but Peru Margin sites 1227 and 1229, and the upper clay layer (7.5 mbsf) from the Sea of Okhotsk are dominated by Chloroflexi (Inagaki et al., 2003, 2006; Parkes et al., 2005; Webster et al., 2006a).

The exact taxonomic associations of the phylotypes in the 16S rRNA gene libraries can be seen from phylogenetic trees published in the papers cited in Fig. 1. These show that many of the Proteobacteria and some of the Gram positive bacterial phylotypes are related to cultured species. Examples of their genera include Ralstonia, Comamonas, Halomonas, Pseudomonas, Acinetobacter, Pedomicrobium, Sphingomonas and Pelobacter from the Proteobacteria and Actinomyces, Clavibacter and Arthrobacter (Actinobacteria) and Bacillus (Firmicutes) from the Gram-positive Bacteria (e.g. Rochelle et al., 1994; Inagaki et al., 2006; Webster et al., 2006a). However, the phylotypes from other subsurface bacterial phyla are not closely related to known cultured species.

The extremely diverse but poorly characterized phylum Chloroflexi is a widespread group of bacteria found in a range of microbial communities, not only subsurface sediments (Coolen et al., 2002; Webster et al., 2006a) but also hot springs, hydrothermal sediments, soils, wastewater and polluted sites (Hugenholtz et al., 1998; Sekiguchi et al., 1999; Teske et al., 2002). Currently the Chloroflexi are divided into at least five subphyla (Hugenholtz et al., 1998; Hugenholtz & Stackebrandt, 2004) with a small number of representative cultured species belonging to subphyla I, II and III. Deep subsurface clones mainly fall into subphyla II and IV, although this is likely to change as new sequences are retrieved, since subsurface sediments often contain a number of Chloroflexi with large sequence diversity belonging to different subphyla. For example, in one study (Inagaki et al., 2006) Chloroflexi-related sequences belonged to three subphyla and one deep branching unclassified group, with individual clones showing up to about 30% sequence difference. Conversely, candidate division JS1 clones at present all belong to one monophyletic group with limited sequence variation (Webster et al., 2004), although greater sequence variation is suggested as a number of different JS1 phylotypes can be recognized by JS1-targeted small 16S rRNA gene fragment PCR-DGGE (Webster et al., 2007). The environmental distribution of JS1 does not seem to be as widespread as Chloroflexi with JS1 phylotypes seemingly being restricted to anoxic sedimentary habitats (Webster et al., 2004, 2007). Within these two subsurface bacterial phyla there are no cultured members of JS1 and few cultured members of the Chloroflexi, and so the physiology and ecological role of these bacteria in the deep subseafloor biosphere is difficult to predict. The closest matching cultured representative of the Chloroflexi to many deep marine biosphere sequences is the H2-dependent dehalorespiring bacterium Dehalococcoides ethenogenes (Maymó-Gatell et al., 1997) with a sequence similarity of about 89%. For this reason and the ability of other cultured Chloroflexi to utilize/produce hydrogen (Sekiguchi et al., 2003; Madigan & Martinko, 2006; He et al., 2007) it has been speculated that subsurface Chloroflexi may also consume hydrogen and anaerobically degrade recalcitrant carbon sources (Wilms et al., 2006a).

Recently, investigations have considered the lithological and/or geochemical conditions that have led to the abundance of Chloroflexi and JS1 in subsurface sediments. One study on the sediments from the Sea of Okhotsk found profound differences in the bacterial composition between different sediment lithologies. In this study pelagic clay layers were dominated by Chloroflexi and JS1, and volcanic ash layers were dominated by Gammaproteobacteria (Inagaki et al., 2003; Fig. 1). However, clear indications of similar differences in community composition associated with sediment lithology are not apparent from other studies (Fig. 1). It has also been stated that JS1 dominate methane hydrate bearing sites and Chloroflexi dominate organic-rich subseafloor sediments (Inagaki et al., 2006). Although, we cannot substantiate either this or other overall differences in community composition due to lithology or geochemical composition with the data in Fig. 1. In addition, JS1 and Chloroflexi have also been detected in shallow subsurface sediments (below 2 m; Webster et al., 2007). In this environment it has been suggested that JS1 dominate strictly anoxic organic-rich, but poor quality, recalcitrant carbon muddy sediments with low sulphate concentrations and small pore size, and are outcompeted by Chloroflexi in subsurface sandy sediments.

Sulphate reduction is an important geochemical activity in the deep marine subsurface (Parkes et al., 2000; D'Hondt et al., 2004) and therefore, sulphate-reducing bacteria (SRB) would be expected to be among the dominant physiological groups. However, only limited numbers of phylotypes belonging to Deltaproteobacteria have been isolated from this habitat (Kormas et al., 2003), and most are only distantly related to known deltaproteobacterial SRB (e.g. Inagaki et al., 2006; Webster et al., 2006a). It has been proposed by Parkes (2005), from calculations of SRB numbers using specific sulphate reduction rates at ODP site 1229, that SRB are below detection in subsurface sediments by PCR using general bacterial 16S rRNA gene primers and are, therefore, absent from gene libraries.

One study reported the use of the specific functional gene dsrA (dissimilatory sulphite reductase) to target SRB in deep subsurface sediments of the Peru Margin ODP sites 1228 and 1229 (Webster et al., 2006a). From this work it was suggested that very low numbers of uncultured SRB must be present at these sites, as only one sediment depth at site 1228 showed the presence of detectable dsrA phylotypes [Webster et al., 2006a; related to uncultured sequences from hydrothermal sediments (Dhillon et al., 2003)]. Similarly, quantitative real-time PCR (Q-PCR) on deep sediments from the Peru Margin ODP site 1227 to determine the relative abundance of dsrA genes (Schippers & Neretin, 2006), showed that DNA copy numbers of dsrA were about 10–3000-fold less than for the 16S rRNA genes of Bacteria or prokaryotes and had an irregular depth distribution.

The above studies may indicate that the subsurface SRB population is very small but active over geological timescales or, alternatively, that sulphate reduction is carried out by unknown sulphate-reducing prokaryotes with divergent functional genes that are not detected using current PCR methods. The data of Mauclaire (2004), in marked contrast to the above, demonstrated that SRB were a major population throughout sediments of Peru Margin site 1229 [constituting 6–22% of all prokaryotic cells using catalysed reporter deposition-fluorescence in situ hybridisation (CARD-FISH)] and at some depths (1, 3.5 and 110 mbsf) represented all detected bacterial cells. It should be noted that the SRB probes used in this study also target other Bacteria including Chloroflexi, which are a major component of the bacterial community within Peru Margin sediments (Fig. 1).

Community composition of Archaea

Reliable identification of new subsurface archaeal 16S rRNA gene clones is essential for understanding the distribution of these microorganisms in the deep subseafloor biosphere, especially because subsurface archaeal populations comprise almost exclusively of uncultured lineages. Unfortunately, many archaeal sequences in molecular diversity studies of marine sediments are often left unidentified or assigned to groupings used in the specific publication (e.g. Fang et al., 2005; Kim et al., 2005; Wang et al., 2005; Heijs et al., 2007), resulting in some confusion. For this reason, we have adopted the subsurface Archaea naming system recently reviewed by Teske & Sørensen (2008). Figure 2 shows the composition of the archaeal populations from 47 16S rRNA gene libraries from 11 published studies of the deep marine biosphere from the Pacific Ocean and one recent study of deep subsurface sediments from the Atlantic Ocean (down to 1626 mbsf; Roussel et al., 2008). The Crenarchaeota (six groups) dominated the Archaea, with 73.4% of the clones, while only 24.5% of the clones belonged to the Euryarchaeota (eight groups). The most abundant archaeal groupings overall were the crenarchaeotal groups, Miscellaneous Crenarchaeotic Group (MCG) and Marine Benthic Group B (MBG-B; synonymous with the Deep-Sea Archaeal Group, DSAG; Inagaki et al., 2003), comprising 33% and 26.3% of clones, respectively. The next most abundant groups were the Marine Group 1 (8.4%; Crenarchaeota), the South African Gold Mine Groups (SAGMEG) 1 and 2 (7.6%; Euryarchaeota) and the thermophilic Euryarchaeota (7.6%); none of the other groupings accounted for more than about 4.5% overall. These groups have also been found in other sedimentary, aquatic and terrestrial environments, and so are not confined to the deep marine biosphere (Teske, 2006a, b). The only Archaea phylotypes closely related to cultured species were the euryarchaeotal methanogens, thermophiles and hyperthermophiles. However, these only accounted for <8% of the clones and were not a major component of the phylotypes; thus, as with the Bacteria, most of the Archaea were from uncultured lineages.

2

Community composition of Archaea 16S rRNA genes from various sites and depths in the deep subseafloor biosphere. See Fig. 1 for details of the diagram construction. The citation codes are: A, Reed (2002); B, Inagaki (2003); C, Kormas (2003); D, Newberry (2004); E, Sørensen (2004); F, Parkes (2005); Webster (2006a); G, Biddle (2006); H, Teske (2006a; & pers. commun.); I, Inagaki (2006); J, Sørensen & Teske (2006); K, Roussel (2008). As far as possible, the nomenclatures of subsurface unclassified archaeal groups are based on their first reported use and as reviewed by Teske & Sørensen (2008).

Most investigations of archaeal diversity in the marine subsurface have used PCR amplification of extracted sediment DNA, although Sørensen & Teske (2006) used extracted RNA and reverse transcription (RT)-PCR to make clone libraries. These authors found mainly MBG-B Archaea in the sulphate–methane transition zone (SMTZ) of ODP Peru Margin site 1227 and mainly MCG elsewhere in the core, but no methanogens despite high methane in the sediment at depth. However, it is unlikely that the MBG-B alone are indicative of the SMTZ, as another Peru Margin study (Biddle et al., 2006) found that four of these zones (ODP sites 1227, 1229 and 1230) were dominated by mixtures of MBG-B and MCG Archaea. Because both of these studies used RT-PCR it is likely that these two groups are the most active Archaea in deep subsurface sediment SMTZs. Another example of MBG-B archaeal signatures in different geochemical zones comes from Inagaki (2006) who proposed that methane hydrate containing sediments are dominated by DSAG (MBG-B).

Deep subseafloor biosphere archaeal clone libraries are seemingly less diverse than was seen in those for Bacteria, with 24/47 libraries presented in Fig. 2 containing only one or two of the 14 main archaeal taxa found in the deep marine biosphere. This could be either an artefact, perhaps due to preferential PCR amplification of specific groups of related genes (Webster et al., 2003), or it may reflect real variation between sites and depths. Nevertheless, there is still a substantial diversity of Archaea phylotypes in the samples investigated. For example, the MCG sequences alone show very large sequence variation (up to 20%) in most of the 27 studies that found this group, and similar results have been reported for other major groups of subsurface Archaea. Such large phylogenetic diversity and widespread distribution of MCG Archaea supports the theory that these anaerobes are a metabolically diverse group of microorganisms that utilize complex carbon substrates (Biddle et al., 2006; Webster et al., 2006a; Teske & Sørensen, 2008).

The community composition of methanogens has been specifically investigated because of their inferred importance in biogeochemical processes and in the production of biogenic methane, which might contribute to the methane hydrate reserves in the deep marine biosphere (Waseda, 1998; Parkes et al., 2000; Lanoil et al., 2001; D'Hondt et al., 2004; Inagaki et al., 2006). However, only small numbers of methanogen clones have been detected directly in general archaeal 16S rRNA gene libraries. For example, at Peru Margin ODP site 1229 only 1/103 clones was thought to be from a methanogen (deeply branching within the Methanobacteriales; Parkes et al., 2005). This accounts for the very low average methanogen content shown in Fig. 2 (0.1%, and not visible in the figure). A small number (6/348 prokaryotic phylotypes investigated) of the Thermococcales and hyperthemophiles in Fig. 2 from the Peru and Cascadia Margins (Inagaki et al., 2006) were also methanogens belonging to the Methanosarcinales, Methanobacteriales and Methanococcales, some of which were closely related to cultured methanogens (e.g. Methanococcus aeolicus, Methanoculleus palmaeoli). These results suggest that, like SRB, known methanogens constitute a very small proportion of the prokaryotic community in the deep sub-seafloor biosphere. Targeted studies using taxon-specific 16S rRNA and mcrA (α-methyl coenzyme-M reductase) gene primers have successfully amplified related sequences to Methanosarcina mazei and Methanosarcina barkeri (Methanosarcinales), and Methanobrevibacter arboriphilus (Methanobacteriales) (Marchesi et al., 2001; Newberry et al., 2004; Parkes et al., 2005). Overall, these results suggest limited diversity of methanogens in the deep marine biosphere.

Abundance of Bacteria vs. Archaea

The majority of molecular studies of prokaryotic diversity in the deep subseafloor biosphere have tended to use conventional or nested PCR with primers specific for either Bacteria or Archaea on extracted DNA, making it difficult to compare directly the magnitude of these two Domains within subsurface sediments. Recently, quantitative molecular techniques have been used to address this question, including FISH, CARD-FISH, and Q-PCR. Bacteria were enumerated by CARD-FISH in depth profiles at two Peru Margin (sites 1227, 1230) and two equatorial Pacific sites (1225, 1226) and were found to comprise a high proportion of the total cell count (acridine orange direct count, AODC), but Archaea could not be detected (Schippers et al., 2005). These authors also used Q-PCR on extracted DNA to estimate the 16S rRNA gene copy numbers of total prokaryotes, Bacteria and Archaea, and found that Archaea were 10–1000-fold less abundant than Bacteria, and that prokaryotic and bacterial 16S rRNA gene copy numbers were indistinguishable (Schippers & Neretin, 2006). Similar results were also found by Inagaki (2006) using Q-PCR. These results suggested that Bacteria dominated the prokaryotes in subsurface sediments, with Archaea contributing only a small proportion. In addition, a high proportion of the cells detected by AODC were also shown to be viable. Conversely, one study (Biddle et al., 2006) of samples from ODP Leg 201 sites 1227, 1229 and 1230 using FISH, showed that Archaea were, on average, 82% of the total prokaryotic community. Evidence for these high subsurface archaeal numbers was further supported by high abundance of archaeal intact polar lipids (average=34%Archaea; Biddle et al., 2006). These conflicting results show that more work on the refinement of molecular methodologies is needed before the true relative abundance of Bacteria and Archaea can be determined in the deep marine biosphere.

Cultured biodiversity

Studies involving laboratory cultivation have been used to examine the diversity of culturable prokaryotes in the deep marine biosphere since the late 1980s, and many of these have been aimed at isolating pure cultures of typical deep sediment organisms. Some of the earliest studies used most probable number (MPN) techniques to enrich and count viable anaerobic heterotrophs, ammonifiers, acetogens, sulphate reducers, methanogens and aerobic ammonifiers in the Peru Margin (ODP sites 680, 681) and Japan Sea (ODP site 798) (Cragg et al., 1990, 1992). Generally the profiles of anaerobes followed those of the relevant geochemical and activity profiles, with highest counts near the surface and elevated counts coinciding with deeper peaks of activity. However, the aerobic ammonifier counts were 100–106-fold higher than anaerobic counts with little decrease with depth, suggesting that these heterotrophs were probably facultative anaerobes surviving in this anoxic habitat.

These studies have led to isolations of a few well-described novel prokaryotes with some physiological characteristics that suit the isolates for growth in the deep biosphere. These pure cultures include the barophilic SRB Desulfovibrio profundus (optimum activity at in situ pressures with activity up to 400 bar; Bale et al., 1997) from the Japan Sea, the methanogen Methanoculleus submarinus (Mikucki et al., 2003) from the Nankai Trough and thermophilic Firmicutes in the genus Thermosediminibacter (Lee et al., 2005) from Peru Margin sediments (sites 1227, 1228, 1230). Despite these prokaryotes having similar physiologies to Bacteria and Archaea represented in deep sediment clone libraries, only M. submarinus has any close sequence similarity (97%) to a clone from the subsurface (Cascadia Margin clone ODP1251A5.5; Inagaki et al., 2006).

Other studies have also isolated pure cultures from the marine subsurface that are less well described. Toffin (2004) screened anaerobic enrichments designed to grow heterotrophs, acetogens and SRB from the Nankai Trough (ODP site 1173) and found mainly Firmicutes, Gamma- and Deltaproteobacteria and Spirochaetes, but only two were isolated as pure cultures and these were closely related to existing Marinilactibacillus and Acetobacterium (Firmicutes) species. Another study obtained 168 isolates from three Equatorial Pacific sites and four Peru Margin sites from Leg 201 (D'Hondt et al., 2004; Batzke et al., 2007) belonging to six distinct lineages. These isolates included Alpha- (26%), Gamma- (18%), Deltaproteobacteria (1%), Firmicutes (44%), Actinobacteria (9%) and Bacteroidetes (1%), with the most abundant being close relatives of the genera Rhizobium, Bacillus and Vibrio (24%, 42% and 14% of all isolates, respectively). Another investigation of Leg 201 samples (Peru Margin site 1230) used aerobic heterotroph and anaerobic methanogen enrichments (Biddle et al., 2005a). The methanogen enrichments failed to yield any methanogens, even after using the same techniques used previously to isolate M. submarinus (Mikucki et al., 2003), but the aerobic heterotroph enrichments gave six pure cultures of Gammaproteobacteria closely related (98–99% 16S rRNA gene sequence similarity) to the genera Photobacterium, Halomonas, Shewanella and Vibrio. As described earlier, Gammaproteobacteria are common in 16S rRNA gene libraries from deep subsurface sediments (Fig. 1), and the genera isolated are often closely related facultative anaerobes with physiologies that suit them to life in the deep subseafloor biosphere (see Biddle et al., 2005a, b; Teske, 2006b; Batzke et al., 2007). Gammaproteobacteria belonging to Halomonas and Marinobacter have also been isolated from the Sea of Okhotsk (Inagaki et al., 2003) and Peru Margin ODP sites 1228 and 1229 (Biddle et al., 2005b). Interestingly, the isolation of facultative anaerobic genera, both aerobically (Biddle et al., 2005a, b) and anaerobically (D'Hondt et al., 2004; Batzke et al., 2007), supports the earlier findings that many of the heterotrophic Bacteria in the deep subsurface may be facultative anaerobes (Cragg et al., 1990).

However, despite a large number of the cultured prokaryotes isolated from deep subsurface sediments falling within higher taxa commonly obtained by cultivation-independent approaches (e.g. Gammaproteobacteria), they are very rarely representative of the abundant largely uncultivated phylotypes. This low culturability of subsurface prokaryotes clearly indicates that a lot more work is needed to develop new strategies for the isolation of deep biosphere prokaryotes, including at elevated pressure, which will undoubtedly reveal novel modes of metabolism.

Comparison of deep biosphere and near-surface sediment community composition

Recently, several studies have examined prokaryotic biodiversity in both marine surface sediments and the shallow subsurface, mainly by molecular methods, at depths down to around 6 mbsf, allowing comparison of these two adjacent habitats. Bacteria in cores from the Skagerrak, German Wadden Sea tidal-flats and the Benguela Upwelling System show that, overall, the upper layers are dominated by Gamma- and Deltaproteobacteria, and the deeper layers by Chloroflexi and candidate division JS1 (Wilms et al., 2006a, b; Parkes et al., 2007a; Schäfer et al., 2007; Webster et al., 2007). It is also striking that in these sediments more SRB phylotypes were identified by 16S rRNA and dsrA genes, with similar results being found for methanogens using 16S rRNA and mcrA genes (Wilms et al., 2006b, 2007; Parkes et al., 2007a). In another study on the South China Sea the Bacteria were dominated by Gammaproteobacteria and the Archaea, similar to the deep subsurface, were dominated by MBG-B, MCG and uncultured Euryarchaeota (Jiang et al., 2007). Overall, these studies show a transition from the near surface layers, where methanogens and SRB are easily detectable, to a deeper subsurface population below where sulphate reduction and methanogenesis still occurs, but the responsible prokaryotes are difficult to detect.

Prokaryotic activity

Methods

Activity estimates have predominantly been undertaken anaerobically using replicate syringe mini-cores subsampled from intact whole round cores and injected evenly with radiotracer substrates (Parkes et al., 1995). Radiolabelled reaction products are then assessed after incubation at in situ temperature for varying incubation times, designed to give measurable product yields. For example, estimates of sulphate reduction use 35SO42− as a radiotracer and measure 35S in the sulphide produced, methanogenesis measures the transformation of 14C-labelled substrates, such as 14C-acetate or 14C-CO2 to 14CH4, and heterotrophic growth is estimated by the incorporation of 3H-thymidine into DNA. It should be noted that these activities are always referred to as potential rates because it is impossible to measure activity without disturbance of the deep biosphere. However, these potential rates are believed to be reliable (Parkes et al., 2000), because they: (1) correlate with geochemical and sedimentological changes; (2) correspond with stable isotopic values of reactants/products; (3) compare well with long-term sulphate removal rates in laboratory experiments incubated anaerobically at 4 °C. The details of these methods have been described and discussed elsewhere (e.g. Cragg et al., 1992; Wellsbury et al., 1993, 2002; Parkes et al., 2000; Kallmeyer et al., 2004).

Depth profiles of activity

Rates of many activities in the deep biosphere (Fig. 3) are usually highest in the upper layers and decrease thereafter, especially at low activity sites which have low oceanic carbon flux into the sediment, and so AODC counts are low (e.g. D'Hondt et al., 2004; Parkes et al., 2005). Decreasing activity with depth occurs for sulphate reduction at Blake Ridge in the Western North Atlantic (Fig. 3ci), in the Equatorial Pacific (site 1226; Parkes et al., 2005) and Woodlark basin, near Papua New Guinea (ODP Leg 180, sites 1109 and 1115; Wellsbury et al., 2002). Similar profiles are seen for thymidine incorporation in the Equatorial Pacific ODP Leg 201 sites 1225 (Fig. 3ai) and the Japan Sea ODP Leg 128 site 798 (Cragg et al., 1992). Rates for methanogenesis normally peak just below the region of highest sulphate reduction activity due to reduced competition from SRB, this can be seen for H2 : CO2 methanogenesis in Equatorial Pacific sites 1225 (Fig. 3aii), the Peru Margin site 1229 (Fig. 3bii; Parkes et al., 2005) and Woodlark Basin (ODP Leg 180, site 1109; Wellsbury et al., 2002).

3

Selected depth profiles of various potential activity estimates from the deep marine biosphere. (ai–ii) Profiles from a low activity site; Equatorial Pacific, ODP Leg 201, site 1225 (unpublished data). (bi–iii) Profiles from a high activity site; Peru Margin, ODP Leg 201, site 1229 (replotted from data in Parkes et al., 2005). (ci–vi) Profiles from a methane hydrate site; Blake Ridge, ODP Leg 164, site 995 (redrawn from Parkes et al., 2000; Wellsbury et al., 2000). In (b) the shaded horizontal bands show the positions of upper and lower SMTZ at about 30 (25–35) mbsf and 90 (85–95) mbsf. The shaded band in (c) indicates the inferred extent of the methane hydrate stability zone (Wellsbury et al., 2000).

Rates of sulphate reduction and methanogenesis in upper layers of the subsurface marine biosphere are comparable to or below those generally found in anaerobic near surface coastal sediments (Parkes et al., 2000). However, sulphate reduction rates usually decrease to zero much more rapidly than methanogenesis because sulphate reduction consumes its own electron acceptor SO42−, while CO2, H2 and acetate, common substrates for methanogenesis, can be replenished by heterotrophic activity, generation at elevated temperatures, and/or other processes occurring in deeper layers (Wellsbury et al., 1997; D'Hondt et al., 2004; Biddle et al., 2006; Parkes et al., 2007b).

Against this background of decreasing activity with depth there are many cases when specific environmental conditions lead to enhanced activity in deeper layers and some examples of this are outlined below. Firstly, anaerobic oxidation of methane (AOM) often results when methane accumulates in the presence of low concentrations of sulphate (Hoehler et al., 1994). For example, AOM occurs in Blake Ridge subsurface sediments (site 995, Fig. 3cvi), in the region where methane accumulates above the top of the gas hydrate stability zone (GHSZ) and sulphate reduction is very low (Fig. 3ci). Secondly, AOM is also stimulated deeper at the base of/and below the GHSZ, where both acetate and H2 : CO2 based methanogensis increase and methane accumulates (Fig. 3c; Wellsbury et al., 2000). Similar changes have also been observed in the Cascadia Margin (Leg 146, site 889/890; Cragg et al., 1996), again at the base of a methane hydrate zone. The accumulation of acetate at Blake Ridge also resulted in enhanced acetate turnover to CO2 below the methane hydrate stability zone (Fig. 3iii). Thirdly, stimulated AOM has also been inferred from geochemical and microbiological changes at the upper and lower SMTZ in the Peru Margin (site 1229, Fig. 3b; Parkes et al., 2005). The lower SMTZ at site 1229 is due to the deep flow of brine bringing sulphate into the deep methane containing sediments, stimulating AOM linked with sulphate reduction (Fig. 3b; Parkes et al., 2005). Stimulation of activity by deep sulphate flux has also been observed at a number of other subseafloor sites (e.g. Nankai Trough and Eastern Flank of the Juan de Fuca Ridge; Mather & Parkes, 2000; Parkes et al., 2007b; Engelen et al., 2008). Lastly, it is clear that thymidine incorporation is often maintained at significant rates deep into subsurface sediments, as observed in the equatorial Pacific, Peru Basin, Blake Ridge (see Fig. 3ai, bi and cv), and the Woodlark Basin (site 1109; Wellsbury et al., 2002). It should be noted that these observed increases in subsurface activity are often reflected in local peaks of prokaryotic numbers estimated by AODC and MPN counts. For example, AODC values increased about 30-fold immediately below the GHSZ in Blake Ridge, and about 6- and 60-fold at the upper and lower SMTZs, respectively, in the Peru Margin site 1229 (see Fig. 3b).

Relating biodiversity and activity

It is important to take into account geochemical profiles when attempting to relate microbial biodiversity and activity. Before extensive molecular biodiversity data were available it was necessary to draw conclusions about prokaryotic composition from directly obtained activity and geochemical profiles (Cragg et al., 1996; Wellsbury et al., 2000) or from activity estimates derived from geochemical data using geochemical flux models (D'Hondt et al., 2002, 2004). For example, when SO42− concentrations were decreasing rapidly with depth and sulphate reduction was high, SRB were thought to be dominant. Similarly, when methane accumulated and methanogenesis was high, methanogens were assumed to be dominant. Furthermore, when MnO2 or NO3 reduction was predicted to be at its highest rate using flux models, manganese reducers and nitrate reducing heterotrophs would be expected. These views might not have always been clearly enunciated but the implications were often clear, especially when supported by evidence from MPN counts (Cragg et al., 1990, 1992), cultured isolates and later from inferred physiologies of the phylotypes present in clone libraries.

However, recently many contradictions have become apparent and need to be explained. For example, methanogenesis, sulphate reduction and AOM are key processes when either measured directly or inferred from geochemical profiles (see Fig. 3 and previous section). However, phylotypes with high sequence similarity to known SRB and methanogens are rare in 16S rRNA gene libraries, as are the archaeal members (ANME) of communities believed to be active in AOM (Orphan et al., 2001). Nevertheless, some methanogen and ANME phylotypes were enriched in a sealed ODP borehole at the Cascadia Margin site 892 (Lanoil et al., 2005) and recently ANME sequences were found to be the dominant archaeal group in some deep sediments from the Newfoundland Margin (Roussel et al., 2008). This apparent contradiction in the majority of studies presents a dilemma unless these groups are affecting the geochemical profiles by the activity of small-sized communities that do not dominate the total prokaryotic population assessed by AODC. In which case what type of physiologies do the majority of the population exhibit? Unfortunately, the clone libraries do not help to answer this question because nearly all of the archaeal and most of the bacterial phylotypes found are only distantly related to cultured species. Furthermore, it is well known from the 16S rRNA gene phylogenetic trees of cultured species that taxonomic closeness is not a reliable predictor of physiology (Gray & Head, 2001).

To resolve this dilemma scientists have used other approaches to link biodiversity with activity and associated geochemical profiles. One study concentrating on SMTZs at three ODP Leg 201 sites (1227, 1229, 1230; Biddle et al., 2006) used carbon flow reconstructions based on the δ13C content of individual cells identified using FISH, intact polar membrane lipids and sedimentary organic carbon. These authors concluded it was organic carbon compounds, other than methane, that provided the major carbon source for the substantial populations of Archaea (MBG-B and MCG) in sediments at SMTZs. However, Sørensen & Teske (2006) suggested that as both MBG-B and MCG were most active in the SMTZ, based on RT-PCR amplified rRNA, then they must benefit directly or indirectly from AOM, although, as noted above, Biddle (2006) suggested that these Archaea do not incorporate methane-derived carbon. This idea is supported by the incorporation of 13C-acetate by stable isotope probing (SIP; Radajewski et al., 2000) into the DNA of MCG Archaea from tidal sediment slurries incubated under anaerobic conditions (G. Webster, J. Rinna, J.C. Fry, A.J. Weightman & R.J. Parkes, unpublished data). It is also possible that the abundant uncultivated groups of Bacteria are heterotrophic, as many isolates and phylotypes of Gammaproteobacteria from deep sediments are closely related to cultured heterotrophs. SIP has also shown that members of the abundant bacterial candidate division JS1 are able to utilize acetate and glucose or glucose metabolites (Webster et al., 2006b). Furthermore, profiles of thymidine incorporation can show that heterotrophic prokaryotic growth is either maintained at significant rates (Fig. 3cv; Wellsbury et al., 2000) or is maximal (Fig. 3bi; Parkes et al., 2005; Fry et al., 2006) in subsurface sediments. Although, organic matter in deep sediments becomes recalcitrant during burial, this can be counteracted by thermal activation of organic matter at depth, providing deep substrates for prokaryotic growth (Wellsbury et al., 1997; Horsfield et al., 2006; Parkes et al., 2007b). Taken together these results strongly suggest that heterotrophy should be an important process in the deep marine biosphere.

Multivariate statistical methods have been used by small numbers of environmental microbiologists for many years to explore and simplify complex patterns in intricate datasets, and have mainly concentrated on cluster analysis and ordination techniques such as principal component analysis (PCA) and multidimensional scaling (MDS) (Ramette, 2007). Recently, cluster analysis has been used to investigate DGGE profiles of sediment communities down to about 5 mbsf (Wilms et al., 2006a, b; Webster et al., 2007). These studies have effectively shown that community composition is different in distinct sediment depth horizons using 16S rRNA gene profiles (Wilms et al., 2006b; Webster et al., 2007) and in different geochemical layers (e.g. at sulphate and methane peaks) with mcrA and dsrA genes (Wilms et al., 2006a).

A more extensive approach was reported by Fry (2006) using cluster analysis, PCA, MDS, correlation and multiple regression of DGGE profiles of 16S rRNA genes, amplified from extracted DNA, to interrelate community composition with activity and geochemistry at two ODP Leg 201 Peru Margin sites (1228 and 1229). PCA was shown to be the best method to explore the link between the diversity profiles, activity and geochemical variables. The first three PCA components accounted for more of the variability in the DGGE profiles of Bacteria, Archaea, Euryarchaeota and JS1 16S rRNA genes at site 1229 (72–79%) than for site 1228 (54–72%). Furthermore, multiple regression of the first three components with all activity and geochemical variables gave good explanations of the components at both sites (1229, 31–95% explanation with the best regression equations, two to five variables; 1228, 8–100%, two to five variables). This clearly showed that community diversity, with all 16S rRNA gene primers used, was strongly related to both the geochemical environment and the prokaryotic activity in these deep subsurface sediments. Furthermore, it suggests that genomic DNA approaches, like directly extracted rRNA methods (Sørensen & Teske, 2006) are also able to identify active and dynamic prokaryotic populations in the marine subsurface.

Relating biodiversity and activity at Peru Margin site 1229

The highly productive Peru Margin site 1229 from ODP Leg 201 is a good example for this section because it has been studied in more detail than any other deep marine subsurface site (Jørgensen et al., 2006). Furthermore, its microbiology has been studied on two separate ODP cruises and results have proved remarkably consistent as far as comparisons can be made (Cragg et al., 1990; Parkes et al., 1990, 2005).

The Peru Margin has very high phytoplankton productivity, stimulated by nutrient rich upwelling waters from the deep Pacific Ocean, so the surface sediment in this area has high organic matter input (up to 8% TOC). The geochemistry of site 1229 has been well described elsewhere (Shipboard Scientific Party, 2003; D'Hondt et al., 2004; Parkes et al., 2005; Fry et al., 2006), but briefly, is typified by two SMTZs, one at 30 mbsf and a second at 90 mbsf associated with a deep brine incursion. Between these two SMTZs sulphate remains low and methane concentrations are high. This resulted in increased AODC counts at around 30 mbsf and the highest deep subsurface AODC values ever recorded at about 90 mbsf. The deep brine incursion also resulted in high dissolved Mn concentrations in this deeper layer (below 110 mbsf) and there was also elevated Mn between the two SMTZs.

From the opposing sulphate and methane profiles giving two sharp SMTZs it would have been expected to find high populations of SRB and methanogens in the sulphate and methane zones respectively. However, this was not the case; only one 16S rRNA gene sequence related to methanogens was found at 42 mbsf. Similarly, although AOM was not directly measured at this site it was inferred from the methane/sulphate gradients, but ANME sequences were not found (Parkes et al., 2005; Webster et al., 2006a). However, Q-PCR at a SMTZ at another Peru Margin site (1227) showed elevated archaeal DNA and rRNA indicating active Archaea in this zone (Sørensen & Teske, 2006). This was confirmed with archaeal reverse transcription 16S rRNA gene-derived (dominated by MBG-B and MCG) clone libraries at site 1229 SMTZs (Biddle et al., 2006). At site 1229 there was also high heterotrophic thymidine incorporation between about 10–90 mbsf (Fig. 3bi), predominance of heterotrophic Gammaproteobacteria in bacterial clone libraries (Fig. 1), and also an abundance of MBG-B and MCG Archaea (Fig. 2), which are also thought to be heterotrophic (Biddle et al., 2006) in this zone. It is, therefore, possible that heterotrophy dominates the metabolism and the prokaryotic populations in site 1229 above the 90 mbsf SMTZ. In addition, there was active methanogenesis and sulphate reduction at this site (Fig. 3ii and iii), especially at the lower SMTZ. It is also possible that the low numbers of prokaryotes responsible for these processes could be influencing the geochemical profiles over long periods of time due to low utilization of methane and the steady use of sulphate diffusing upwards from the brine incursion, stimulating sulphate reduction at depth (Fig. 3biii). The very low maintenance energy requirement and turnover times calculated for prokaryotes at this site help confirm this hypothesis (Biddle et al., 2006). Further, by analogy with rates of metal reduction calculated at other Leg 201 sites (D'Hondt et al., 2004), it is likely that MnO2 reduction is important below and just above the 90 mbsf SMTZ. It is important to note that the above hypotheses, arising from comparing community composition activity and geochemistry at site 1229, are fully supported by the multivariate analyses of DGGE profiles discussed in the previous section.

Prospects for the future

Research over the last 20 years has increased understanding of the microbiology of the deep marine biosphere enormously. It is likely that the tools currently available and impending advances in molecular approaches will enable understanding of the way this complex habitat works to be increased even more over the next decade. Below we will outline some pointers for future studies we believe will be important (see D'Hondt et al., 2007).

Direct quantitative evidence of the numbers and types of prokaryotes carrying out different functions need to be addressed urgently. Techniques such as CARD-FISH and Q-PCR need to be extended and applied to a range of functional genes and group specific bacterial and archaeal 16S rRNA genes (Schippers & Neretin, 2006), and importantly ground truthed by comparison with geochemical and activity data. However, research over several years has shown that currently used PCR primers are biased against many prokaryotes present in the environment (Baker et al., 2003), so design of new more effective PCR primers is going to be important (Teske & Sørensen, 2008). Recent evidence from the deep sea, using 16S rRNA gene tag 454 pyrosequencing, points to a much greater diversity than demonstrated previously using PCR and cloning (Sogin et al., 2006). Thus, the application of high throughput, cloning-independent sequencing techniques that can produce very large numbers of ≥400-bp sequence reads (Schuster, 2008) will be especially important, and will enable large metagenomic libraries to be analysed, avoiding limitations associated with PCR (Biddle et al., 2007; Wommack et al., 2008). In addition, inherent problems of extracting low concentrations of nucleic acids from low biomass deep subseafloor samples will also be enhanced using methods such as whole genome amplification (Abulencia et al., 2006) to improve DNA yields. Sequencing of metagenomic libraries from the deep subsurface will produce very large amounts of information so enhancements in the software used to analyse this data and computer storage to allow its distribution will also be vital (Hugenholtz, 2007). Such techniques are difficult to apply equally to all groups of prokaryotes and so initially concentrating studies on organic-rich sites with high AODC may give the best chance of success. To allow a full understanding of the deep subseafloor biosphere more multidisciplinary studies will be needed. International cooperation, as demonstrated so successfully for ODP Leg 201, will be essential so that data from biodiversity, activity and geochemistry studies can be combined and the complex datasets analysed by appropriate multivariate analyses. Once all these techniques are working well together it will be essential to concentrate studies with a finer spatial resolution on two or three contrasting deep subsurface sites >1000 mbsf (Roussel et al., 2008) in another microbiology focused deep marine biosphere expedition, similar to ODP Leg 201.

Acknowledgements

We are grateful for the help provided by all the crew and scientific parties on the ODP cruises reported here. We also thank all the postdoctoral scientists who have worked in our research groups on deep marine biosphere projects, without whom writing this review would not have been possible. We thank Dr Andreas Teske for providing us with unpublished biodiversity data from ODP Leg 201, site 1225. Aspects of the work presented in this review were funded by the European Union DeepBUG project (contract number EVK3-CT-1999-00017) and by the Natural Environment Research Council (NERC) Marine and Freshwater Microbial Biodiversity programme (NER/T/S/2000/636 & 2002/00593).

Footnotes

  • Editor: Patricia Sobecky

References

View Abstract